Jump to content

Active, DSP, REW, Xilica 101


rplace

Recommended Posts

43 minutes ago, rplace said:

And here is it with nothing added or subtracted just raw sweep 20-20K for the HF driver only:

 

What does it look like with the HF driver removed from the baffle? I don't see any signs of baffle-edge reflections, but you never know.

 

Otherwise, is that a whizzer cone on the HF driver? If so, it might be the effect of the mechanical crossover from the main cone to the whizzer.

 

The oddest part is that nothing obvious shows up in the frequency response magnitude or spectrogram.

Edited by Edgar
Clarification
Link to comment
Share on other sites

30 minutes ago, Edgar said:

 

What does it look like with the HF driver removed from the baffle? I don't see any signs of baffle-edge reflections, but you never know.

 

Otherwise, is that a whizzer cone on the HF driver? If so, it might be the effect of the mechanical crossover from the main cone to the whizzer.

 

The oddest part is that nothing obvious shows up in the frequency response magnitude or spectrogram.

 

I'll try that today. Work is getting the way, so it might be a bit.

Link to comment
Share on other sites

1 hour ago, rplace said:

Here are the HF sweeps only, no LF. It seems to be there with and without any processing applied...

 

...And here is it with nothing added or subtracted just raw sweep 20-20K for the HF driver only:

 

1069209254_PAPNothingAddedorSubtracted.jpg.97680cf8540629d98d99fec1b09cfcb0.jpg

That's a function of the driver itself: a full-range direct radiating lightweight cone--(e.g., Fostex, etc.).  Apparently wizzer cones really give REW a hard time when it comes time to calculate phase and the derivative of phase: group delay.  If you look at the frequency range of those phase wraps, it's occurring over the 3-4.5 kHz band.  That's also the point where there is apparently a handover of the motion of the cone as a somewhat rigid body to the wizzer cone area only, so you're seeing a broadening of the spectrogram just below 4 kHz--and a very complex behavior.  If you change the resolution of the spectrogram to Fourier instead of wavelet, you can see the complex behavior most clearly in the zigzagging peak energy time curve.  This is a function of that sort of driver, apparently.  Can the ear hear that?  I'm not sure, but apparently not.  But it is the point where the wizzer cone basically breaks free of the rest of the cone and begins to carry the load higher in frequency.

 

Chris

  • Thanks 1
Link to comment
Share on other sites

1 hour ago, rplace said:

Here are the HF sweeps only, no LF. It seems to be there with and without any processing applied

 

With all the PEQs, 1.5ms delay and -7db Gain:

 

PAPHfOnlyWithPEQ.jpg.fbb5f715c2cad85537afe3c1dfb6f81f.jpg

 

 

And here is it with nothing added or subtracted just raw sweep 20-20K for the HF driver only:

 

1069209254_PAPNothingAddedorSubtracted.jpg.97680cf8540629d98d99fec1b09cfcb0.jpg

 

 

 

 

Just looking at the blue graph above, I would say you need to invert the polarity on your HF driver, which correlates with the GD anomaly you posted above that.  Give that a try and see what you get.

Link to comment
Share on other sites

1 minute ago, Rudy81 said:

 

Just looking at the blue graph above, I would say you need to invert the polarity on your HF driver, which correlates with the GD anomaly you posted above that.  Give that a try and see what you get.

 

Rudy, look at my Xilica posts. I'll copy it here so you don't have to go back a page, lower left circled in red. I have it inverted.

 

If I don't invert it I get a big GD spike more around the crossover region, along with this one. If I invert the LF the GD problem goes away but I get one more wrap of phase. If I invert both the big GD spike is back. When I invert the HF only I get the best GD and no phase wraps....beyond my 3-4K problem.  @Chris A please confirm this is the correct way to attack a big jump in GD around the XO point.

 

PAPHF.thumb.JPG.4070676e45d318352bee78419ee1d6f4.JPG

Link to comment
Share on other sites

13 minutes ago, Chris A said:

Apparently wizzer cones really give REW a hard time when it comes time to calculate phase and the derivative of phase: group delay. 

 

So nothing for me to worry about or have the ability to counteract? Is this the case where if it sounds good it is good? If we are assigning blame is it REW or my driver ?

 

This driver (Voxativ 1.6) is quite a bit more expensive than my Tang Bands or my Audio Nirvana. Frankly on the surface I think they are worth it. They sound really good right out of the box. After an hour of EQ and delay even better. Neither the TBand or ANirvana seemed to show this behavior, both have wizzer cones....but they were also both in a big horn, not Open Baffle. I'll eventually swap them all around given a try for each driver in each set up.

 

I think the reviews of Pure Audio Project Trio might be right.

Link to comment
Share on other sites

For those of you trying to read and learn. I'll try and explain what Rudy and I were talking about. I only know this from Rudy and Chris and their generous amount of time spent with me.

 

If you are moving along with what Chris has so well documented and you see a huge spike in Group Delay around you crossover point don't spends days/hours like I've don't trying to fix it with PEQ. Try inverting the polarity of ONE of your drivers. I always try both one at a time and see which yields better results. For me it happens to always be the HF driver. Not sure why.

 

Check out the previous page to see a much better GD and Phase plots where I have inverted the polarity of the HF driver and things look a lot better.

 

Here is GD plot with a large spike around 500 to 800. I happen to be crossing at or about 500. So it tips me off that polarity is the issue.

1672570002_PAPnormalpolarityonHFGD.jpg.1d7021bbd75e685f5bfad6b30d73bd66.jpg

 

 

And here is the phase/SPL for the same

1524904663_PAPnormalpolarityonHF.jpg.df9d91be868227ab84871b20603573a9.jpg

 

 

both look pretty bad. Here they are with the inversion

2105766136_PAPCloneGD.jpg.88913c6a97073c13d4abd7e4096d95ff.jpgPAPPhase.jpg.1e31e4f97f4f1dcf951438c3e3cd899f.jpg

 

Link to comment
Share on other sites

38 minutes ago, rplace said:

If we are assigning blame is it REW or my driver?

Yes.  I'm not certain, but I believe that REW uses the peak energy time from the measurements, then likely calculates the phase based on the relative time buckets in the stored peak energy time vector.  Then it calculates a time-based derivative on the phase curve that it generated to produce the group delay curve. (This is all speculation as to how REW algorithms work). It could also be a cross-correlation from the stored input signal to calculate phase, with similarly noisy input signal for phase.

 

Derivative calculations are always noisy numerically;  that's why we see step functions used over impulse response because integrating the curve is just like smoothing the curve. 

__________________________________________________________________________________________

These are the same issues that for me goes back to 1980 with a developmental data acquisition system that I used to take measurements of ground shaker vehicles: Vibroseis.  These issues above were issues then, too. 

 

(Very off-topic): A scan of a sales brochure from 1980 of that particular company (which no longer exists..that was 40 years ago) and a vibrator buggie on the test pad constructed for engineering tests, with that data acquisition system keyboard shown:

 

scan007.jpg.ff49b70d73143d1554fdd82a70f4a5ab.jpg

 

Chris

  • Thanks 1
Link to comment
Share on other sites

8 minutes ago, rplace said:

Here is GD plot with a large spike around 500 to 800. I happen to be crossing at or about 500. So it tips me off that polarity is the issue.

Are you saying that If one has a spike or dip around the crossover point that there may be a polarity issue?

Link to comment
Share on other sites

If the driver has a broad "peak energy time", one peak occurs, then another appears a few microseconds later.  Because more than one area of the driver diaphragm is radiating but the two areas are separated in time or distance from the measurement device slightly, then you get the type of measurement behavior that you show at 3-4.5 kHz. 

 

Chris

Link to comment
Share on other sites

3 minutes ago, The Dude said:

Are you saying that If one has a spike or dip around the crossover point that there may be a polarity issue?

 

In my experience, if you note a large dip at the xover point, after setting the timing, the GD will confirm the need for polarity inversion of one of the drivers.  The GD shows a huge change in GD at that point.  Of course, take my comments with a grain of salt as I am merely a student of the magic arts of audio and REW.

  • Like 2
  • Haha 1
Link to comment
Share on other sites

9 minutes ago, The Dude said:

Are you saying that If one has a spike or dip around the crossover point that there may be a polarity issue?

 

Yes, more or less. I believe it is always a spike though.

 

If you start with individual measurements and they look decent then when you combine them there is a spike in GD around the XO point you should try inverting the polarity. I'm sure Chris can give us the more technical explanation. I'm just following the recipe on the side of the box. If this then do that!

 

Also, not just a bump of 2-5ms but an off the chart spike like you see above in my green plot from 500-800. You can also see the drop in SPL around that spot. Not sure but I believe the HF and LF are canceling each other out. If you keep trying to boost in that dipped area it just messes things up in other spots. Take the time to really read Chris' steps on flattening, gain setting, time aligning and it will some day get easier. It is a lot at one time, but if you bite it off in chunks you eventually get there.

Link to comment
Share on other sites

Yes, it's easy to see polarity reversals at the crossover points.  That's usually the way I know that I have issues--by looking at the phase or group delay curves, and seeing phase wraps and a pole in the group delay response.

 

Chris

  • Like 1
Link to comment
Share on other sites

Say when I measure, the mid/hi horn is 5 db hotter than the bass bin, meaning 80 db vs 75 db. Do I cut the gain of the output on that component at the crossover to bring it closer to the 75 db of the bass bin?

Sent from my SM-G955U using Tapatalk

Link to comment
Share on other sites

Yes.  In your example, I'd set the channel gain of the midrange down -5 dB relative to the bass bin.  I usually try to leave the bass bin at zero gain because it usually the least sensitive of the drivers, then set the other channel gains to compensate for the differences in SPL.

 

Chris

  • Like 2
Link to comment
Share on other sites

  • 2 weeks later...

I would just like to say a very special thank you to Chris A for helping on a journey that has produced the most incredible sound from my Klipsch La Scala II's, the man is a true gentleman and has the patience of a Saint 

 

I was very nervous about disconnecting my drivers from the crossover and had apprehension regarding dialing in the Minidsp 10x10hd, I had previously tried Dirac live and was not impressed by the change 

 

However having the offer from Chris A to help dial everything in and accurately time aligne and set all output channels and EQ settings was a blessing 

 

Without his knowledge I would never ever have taken the route of active crossovers 

 

Chris, thank you again your knowledge and help will never be forgotten you have helped me taken my listening pleasure to a complete new level of clarity and emotion 

 

Deano

  • Like 4
Link to comment
Share on other sites

  • 4 weeks later...

Hi All,

So after experiencing what the minidsp could do, I went ahead and pulled the trigger and purchased a Xilica XP4080, which finally arrived today and with Chris A's help I'm already starting to dial in the unit, 

 

I will report back hopefully soon on the differences if any, and thanks to Chris and Dave A for the recommendation, I'm sure I will appreciate the upgrade 

 

Cheers 

Deano 

IMG_20200306_184527.jpg

  • Like 1
Link to comment
Share on other sites

  • 4 weeks later...
  • 6 months later...

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...